|Publication number||US7996639 B2|
|Application number||US 12/750,812|
|Publication date||Aug 9, 2011|
|Priority date||Sep 29, 2004|
|Also published as||US7062624, US7155593, US7281109, US7716441, US20060069862, US20060195659, US20070061515, US20080222355, US20100185828|
|Publication number||12750812, 750812, US 7996639 B2, US 7996639B2, US-B2-7996639, US7996639 B2, US7996639B2|
|Original Assignee||Hitachi, Ltd.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (13), Classifications (19), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The application is a continuation of U.S. patent application Ser. No. 11/900,440, filed Sep. 11, 2007 (now U.S. Pat. No. 7,716,441), which is a continuation of U.S. patent application Ser. No. 11/599,494, filed on Nov. 13, 2006 (now U.S. Pat. No. 7,281,109), which is a continuation of U.S. patent application Ser. No. 11/415,592, filed on May 1, 2006 (now U.S. Pat. No. 7,155,593), which is a continuation of U.S. patent application Ser. No. 10/954,385, filed on Sep. 29, 2004 (now U.S. Pat. No. 7,062,624), the entire disclosure of which is incorporated herein by reference.
The present invention relates generally to data storage systems and in particular to migration of data in a data storage system.
The storage needs of present-day applications have widely varying characteristics. For example, in a database facility, a database application might require high-speed storage to store log files (e.g., redo log files), while the storage of database tables might adequately stored in lower-speed storage. A tiered storage system provides storage volumes having different operational characteristics. Tiered storage systems gives the user (or system administrator) access to a range of performance capability and storage capacity to customize the storage needs for an application. Thus, in the database example, log files might be assigned to a storage volume having high-speed access characteristics. The database tables might be assigned for storage in a lower-tiered storage volume. Tiered storage is especially suitable for managing the cost of storage, by providing flexibility in balancing the changing needs between high speed (expensive) storage and lower performance (low cost) storage in an enterprise.
Data migration must be performed occasionally when the storage capacity of a volume is reached. This involves moving data from the original storage volume to a new storage volume. In a tiered storage system, it might be desirable to maintain the relative tier relationship among the storage volumes associated with an application. High capacity storage systems are increasingly in demand. It is not uncommon that a storage system contains hundreds to several thousands of physical storage devices. Managing such large numbers of physical devices in a tiered storage configuration to effect data migration, where there might be dozens to hundreds of devices at each of a number of tier levels can be a daunting task.
In accordance with embodiments of the present invention provides migration groups can be provided in a tiered storage facility. Operations on the migration group can include movement of the migration group to a different tier position. Data stored in the logical devices associated with the migration group is migrated to target logical devices, the target devices being selected depending on the new tier position. Relative tier position information among the logical devices in a migration group is maintained, thus preserving the tier hierarchy of the migration group.
Aspects, advantages and novel features of the present invention will become apparent from the following description of the invention presented in conjunction with the accompanying drawings, wherein:
The network 8 is in data communication with a storage virtualization system 5. A second communication network 9 connects the storage virtualization system 5 to a plurality of storage subsystems 6. The second communication network 9 can be a SAN.
A management server system 3 is connected to the storage virtualization system 5 over a communication network 10. As shown in
A user such as a system administrator, can be connected to the communication network 10 using a suitably configured console 4 such as a PC (personal computer) or the like. The management server system 3 and the console 4 can communicate using a protocol such as TCP/IP based Ethernet, Token-Ring, FDDI (fibre distributed data interface), and the like.
An example of a storage virtualization system 5 is the MDS 9000 SERIES of multilayer switches manufactured and sold by Cisco Systems, Inc. Another example of a storage virtualization system is an enterprise PC server system based on the IPStorŪ Enterprise Edition software produced and sold by FalconStor Software.
As can be seen in
The storage virtualization system 5 is further configured with one or more management processors for management of the virtualization system. An internal bus interconnects the internal components of the storage virtualization system 5. The management processors perform the processing necessary to present the hosts 2 with a view of a plurality of virtual logical volumes.
The storage subsystem 6 comprises a plurality of storage devices (physical devices) 7 and one or more controllers (e.g., RAID controller) CTL to access the devices. Each controller typically includes a processor, memory, a network interface card (NIC) such as an Ethernet card or an FC port. The controller might include cache memory comprising non-volatile RAM (NVRAM) to protect against power failures. The controller serves to provide ports each having an associated WWN. Thus, in a SCSI environment, the port WWN serves as the target ID; in an FC configuration the WWN serves as the LUN (logical unit number).
Other operational aspects of each storage subsystem 6 might include a module for creating parity information. In the case of a RAID configuration, there may be a parity disk. As will be discussed below, a module is provided to makes accessible over the ports of the storage subsystem 6. A module might be provided for handling SCSI IO operations; i.e., a SCSI interface.
The controller(s) define one or more storage volumes (also referred to as logical units, LUs) among the storage devices 7 that comprise a storage subsystem 6. Thus, each storage subsystem 6 might have one or more logical units defined on it. Communication with the storage subsystems 6 can be based on SCSI-2 or SCSI-3 command sets (small computer system interface).
Though not shown, it can be appreciated that the storage virtualization system 5 may include an administrative component that can be accessed by a console 4. The administrative component allows a system administrator to perform various tasks needed to initialize and otherwise maintain the storage virtualization system. The storage subsystem 6 might be a component of the storage virtualization system 5, or it might be external storage components that connect to the virtualization system.
The management server 3 can provide services including a GUI that can be accessed by the console 4 over LAN 10. The management server includes a tiered volume manager 32. It can be appreciated, however, that the tiered volume manager function can be provided in the storage virtualization subsystem 5. As will be discussed in further detail below, the tiered volume manager 32 defines manages tiered volumes and assigns tiered volumes to the hosts. A group manager 36 allows a user to create migration groups of virtual logical volumes for migration. A migration manager 34 creates, schedules, and otherwise manages migration tasks to effect migration among the tiered volumes.
Each of the logical devices is associated with or otherwise assigned to a tier by a system administrator. The logical devices are referred to in
Column 74 of the tier volume table 70 identifies a list of FREE logical devices that are associated with that tier. For example, the free logical devices in tier 2 include logical devices identified as 110, 111, 112, and 113 to 119. The operational status of these logical devices is that they are not being used by any hosts 2, and thus are not associated with a virtual logical volume. FREE logical devices can be allocated for use, either to serves as storage for a host or for a migration operation (discussed below).
Column 75 of the tier volume table 70 identifies a list of RESERVED logical devices that are associated with that tier. Thus, the RESERVED logical devices for tier 3 include the logical device identified as 120. The operational status of these logical devices is that they are being used in a migration operation. This aspect of the present invention will be discussed further.
The migration manager 34 includes a scheduler for scheduling migration tasks. The migration manager 34 communicates with a migrater 503 which is a functional element in the storage virtualization system 5. The migrater 503 coordinates the migration operation between a source virtual logical volume and a target virtual logical volume. The group manager 36 creates migration tasks that are scheduled by the scheduler component in the migration manager 34. For example, in a UNIX-based OS, the “cron” facility might be employed as the scheduler component.
A typical execution sequence might be:
The storage virtualization subsystem 5 includes a volume manager 501 and, as mentioned above, a migrater 503. The volume manager 501 performs the function of creating (or defining) virtual logical volumes comprising logical devices. A logical device in turn comprises one or more logical units, and a logical unit comprises the storage devices 7 in a storage subsystem 6.
A virtual logical volume is presented to a host 2 via a port; e.g., port 1, port 2, port 3. The volume manager 501 uses two mapping mechanisms to achieve this. First, there is a mapping between the virtual logical volume and the port that is used to access the virtual logical volume. For example, the port can be an FC (fibre channel) port, or a SCSI port. A mapping in the form of a table 610 is shown in
The mapping table 610 also includes a mapping between the virtual logical volume and a logical device. In the mapping table 610, this mapping is shown between the virtual LUN field 613 and an LDEV field 614 which identifies a logical device by way of a logical device number (LDEV).
A system administrator can therefore define/assign virtual logical volumes and host associations by filling in the fields in the table 610. Based on the example shown in
A logical unit within the storage system might be a storage volume that is defined on an entire physical device. A logical unit might be a volume that is defined on only a portion of a physical device, where other portion(s) of the physical device constitute storage volume(s) for other logical unit(s). In the case of a RAID configured volume, the logical unit will typically comprise two or more physical devices. In addition, several logical units may put a same group consisted from RAID configuration. The current RAID standard defines RAID levels 1, 2, 3, 4, and 5. The RAID configured volume can be any one the currently defined RAID levels.
Thus, for example, consider the logical unit 53. The header information in its header area 51 includes an LBA (logical block address) offset on the storage device 7 (
The header information also includes the size of the data portion, and can be expressed in units of number of blocks or number of bytes. The header information identifies the port on the storage device with which the logical unit is associated, and the logical unit number (i.e., the field information from fields 94 and 95 in
The migrater 503 performs migration operations. On-line data stored on a logical device (that is accessed from a host as a virtual logical volume) can be migrated (copied, moved, etc.) to another logical device (the target logical device). The target logical device can be defined in the same storage subsystem 6 as the source logical device, or the target logical device can be a logical device defined on a different storage subsystem 6.
Recall that migration tasks are defined and scheduled by a scheduling process in the migration manager 34. The scheduling process initiates a migration operation automatically according to the schedule, or alternatively, one can be initiated manually.
The following actions occur when a migration task is performed. For purposes of discussion, the source LDEV is the logical device that contains the data to be migrated and the target LDEV is the logical device to which the data is copied. Prior to performing the migration, the source LDEV is actively accessed by a host device; the host device accesses the logical device as a virtual logical volume per the storage virtualization system 5.
The foregoing is referred to as an “on line” migration method. Another migration method involves the use of a bitmap. The LU maintains a bit map for keeping track of changes for mirror and write I/O from host on target LDEV. If bit map is turn on, namely the blocks in target LDEV for the bit map were modified, then the LU reads data from the latest blocks in target LDEV. If bitmap is turn off, namely the blocks in target LDEV for the bitmap were not modified, then the LU reads data from the blocks in source LDEV. This and still other alternative method for migrating data are known.
As described above, the functionality of the tiered volume manager 32 is provided the management server 3, and the migrater 503 is provided in the storage virtualization subsystem 5. It is understood, however, that these functional components can be provided in the hosts. Thus, for example, instead of the virtualization hardware, these features can be incorporated in a PC or server which has at least a processor, memory, network I/Fs (interfaces) and HBAs (host bus adapters) that serve as ingress or egress ports. As another example, the virtualization functionality can be provided via host-based virtualization software such as Veritas VxVM. In such a case, the storage virtualization system 5 does not provide volume management services, but serves only as a fibre channel switch or a fibre channel hub.
Manipulation of the foregoing tables will now be described in connection with creating virtual logical volumes based tiered storage volumes. The administrator defines logical devices as candidate targets for migration, taking into consideration the tiered storage configuration. Thus for a given tier, the administrator might create a set of logical devices of equal capacity. The logical devices are composed of the logical units provided by the storage subsystems 6. Thus, for example, the administrator might define an entry for Tier 1 as shown in
A logical device can then be associated with a virtual logical volume using the fields 613, 614 in table 610 of
The discussion will now turn to creation of migration groups using tiered storage volumes.
When the assignment of virtual logical volumes to a migration group is made, a relative tier position is determined for each in step 303 for each constituent virtual logical volume in a migration group. The tier position of a virtual logical volume is the tier position of its associated logical device. The relative tier position is a tier position metric that is determined relative to the highest tier position among all the virtual logical volumes comprising the migration group. The information is then stored in a step 304 to memory in the management server 3.
Typically, the naming of a virtual logical volume (i.e., it's virtual LUN) is “host-centric”. Thus, each host is likely to have a virtual LUN “1”, a virtual LUN “2”, a virtual LUN “3”, and so on. However, each host will access a virtual LUN over a specific port on the virtual storage system 5. Thus, a host A might access its virtual LUN “1” over port “1” on the virtual storage system 5, while a host B might access its virtual LUN “1” over port “2” on the virtual storage system 5. Thus, in the particular embodiment of the present invention shown in the figures, the port number and the virtual LUN are both needed to uniquely identify a virtual logical volume.
For convenience, the notation “port#:LUN#” can be used to uniquely identify a virtual logical volume. Thus in table 40 of
Using the table of
The relative position field 43 stores the tier positions of the constituent virtual logical volumes in a migration group relative to the “reference tier” in the migration group, referred to as a “tier hierarchy”. An arithmetic subtraction operation can be performed to obtain the relative position. For example, the tier position of virtual logical volume 2:2 in Group A is tier position “1” (because its associated LDEV is 11, which from
I is the relative position (field 43),
M is the reference tier position (field 44), and
T is the tier position of the virtual logical volume.
Thus, the relative position of virtual logical volume 2:2 is “−1”.
It can be appreciated that the relative position can be determined with respect to the lowest tiered volume in the migration group, instead of the highest tier position. In that case, the reference tier field 44 would contain the tier position of the lowest tiered LDEV in the migration group.
It can be appreciated that the reference position can be a tier position that is some value between the highest and lowest tier position in the migration group. In such a case, negative and positive relative position number would be needed to indicate whether a virtual logical volume in the migration group is at a higher or a lower tier position than the reference position.
The discussion will now turn to performing a migration operation in accordance with the present invention.
A tier position area in the window 87 might display the virtual logical volumes that comprise the selected migration group. The example shown in
In accordance with the present invention, movement of a migration group is specified in terms of movement from one tier to another tier. For example, it might be desirable to move the entire migration group “down” one tier position. This means that each constituent virtual logical volume in a migration group would be migrated down one tier position relative to its current tier position. The example shown in
As can be seen in
Alternatively, the window 89 can be a “preview” window used to display the configuration before actually performing the migration. In a storage facility that contains thousands of physical drives, a preview window might be especially helpful in running through “what if” scenarios. Of course, it can be appreciated that the present invention is applicable to storage systems of any size. To complete the discussion of
It can be appreciated that the interfaces shown in
Referring now to
In a step 502, a new reference tier position (field 44,
In steps 503 through 507, the new tier position for each virtual logical volume is computed. The new tier position will be used to select the logical device as the migration target for the data that is stored on the logical device currently associated with the virtual logical volume.
Thus, in step 503, a virtual logical volume is selected from the selected migration group. In step 504, a new tier position is computed for the virtual logical volume. This involves performing the following computation:
T new =M new −I,
Tnew is the new tier position,
Mnew is the newly computed reference tier position, and
I is the relative position (field 43).
This computation is derived from the computation used to determine the relative position above. The specific computation for Tnew of course depends on how the relative position is determined.
If Tnew exceeds the maximum number of tier positions, then it can be set to the highest numbered tier position. Likewise, if Tnew is less than “1” (i.e., the lowest numbered tier position), then it can be set to “1”. Some other convention can be adopted, of course.
In step 505, using the example given above, the table 70 is searched for a logical device in tier position 1 (Tnew). In particular, the FREE field 74 is searched and a free LDEV from among the available tier 1 LDEVs and placed in the RESERVED field 75. This logical device will serve as the target of a subsequent migration operation. In step 506, a migration task is inserted into a task list; this includes identifying the logical device currently associated with the virtual logical volume, either explicitly or by referencing the virtual logical volume, and identifying the target logical device.
In step 507, a determination is made whether there are any more virtual logical volumes to process. If there are more virtual logical volumes to process, then the process returns to step 503.
When all the virtual logical volumes have been assigned to a target logical device, then processing proceeds to a step 508. The task list includes a list of migration tasks for each of the virtual logical volumes in the selected group. The task list is then scheduled for a migration operation at a suitable time, which can be specified by the user or can be automatically scheduled according to some predetermined schedule.
Thus, the notion of moving a migration group to a new tier position (e.g., by specifying a delta value, ΔT) involves migrating each constituent virtual logical volume from its current tier position to a new tier position depending on ΔT. Migrating a virtual logical volume involves migrating the data stored on its associated logical device to a logical device selected from the new tier.
The discussion will now turn to some illustrative examples of operation of the present invention.
The figure also shows the tier positions of the logical devices. The reference tier position (M) for this migration group is “1”. The relative tier position (I) for each virtual logical volume and its logical device is also shown. The convention for discussing tier positions is that the lowered number position represents a “higher” tier in terms of performance. Performance is typically determined based on criteria such as access speed, storage capacity, reliability, and so on; but is not limited to such criteria and might vary from one storage facility to the next.
Now suppose a user such as a system administrator decided to move the migration group to a lowered number tier; i.e., to a lower performance tier position. Thus, the user might want to move the entire migration group down by one tier position; i.e., ΔT=+1.
Suppose a user desires to move the migration group down by two tier positions; i.e., to a lower performance tier; i.e., ΔT=+2.
As for logical device LDEV 4, the migration target logical device should be a logical device in tier position 6. However, tier position 6 is a position that is lower than the lowest tier position, namely, tier position 5. Thus, as discussed in step 504, the migration target logical device will be selected from an available logical device in the lowest existing tier position, namely, tier position 5.
Similarly, the migration target for LDEV C (currently associated with virtual logical volume Vvol 3) is a logical device in tier 4.
Now for the migration target for Vvol 4. According to step 504 the new tier position for Vvol 4 is computed as follows: Tnew=Mnew−I; Mnew is 2 and I is −3, and so the tier position for the migration target logical device is 5. However, since the logical device associated with Vvol 4 is already a tier 5 device, a migration operation is not needed. Moreover, the tier hierarchy has re-expanded, and the relative tier positions among the virtual logical volumes in the migration is restored.
Referring now to FIG. 14E′, another benefit of preserving the relative tier position arises when a new tier position is defined. Thus, starting from the configuration shown in
Thus, in FIG. 14E′ a logical device LDEV E is shown to be available in tier position 6. A user can initiate an operation to expand the tier hierarchy of the migration group. The result is that the data in the logical device LDEV D (associated with virtual logical volume Vvol 4) would be migrated to LDEV E. The virtual logical volume would then be associated with LDEV E.
The examples shown in the figures illustrate “flattening” of the tier hierarchy at the bottom of the tier. However, a similar flattening effect can be observed if the migration group is migrated up to higher performance tier positions. For example, in
M new =M current +ΔT,
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6757778||May 7, 2002||Jun 29, 2004||Veritas Operating Corporation||Storage management system|
|US6779078||Feb 23, 2001||Aug 17, 2004||Hitachi, Ltd.||Data storage system and method of hierarchical control thereof|
|US7222172||Jan 30, 2003||May 22, 2007||Hitachi, Ltd.||Storage system having virtualized resource|
|US20030065898||Sep 9, 2002||Apr 3, 2003||Flamma Bruce M.||System for managing object storage and retrieval in partitioned storage media|
|US20030131182||Jan 9, 2002||Jul 10, 2003||Andiamo Systems||Methods and apparatus for implementing virtualization of storage within a storage area network through a virtual enclosure|
|US20030172149||Jan 23, 2002||Sep 11, 2003||Andiamo Systems, A Delaware Corporation||Methods and apparatus for implementing virtualization of storage within a storage area network|
|US20030221063 *||Sep 5, 2002||Nov 27, 2003||Yoshiaki Eguchi||Method and apparatus for data relocation between storage subsystems|
|US20040083202||Aug 30, 2002||Apr 29, 2004||Arkivio, Inc.||Techniques to control recalls in storage management applications|
|US20040199515||Apr 4, 2003||Oct 7, 2004||Penny Brett A.||Network-attached storage system, device, and method supporting multiple storage device types|
|US20040260875||Jul 23, 2004||Dec 23, 2004||Hitachi, Ltd.||Data storage system and method of hierarchical control thereof|
|US20050010618||Mar 29, 2004||Jan 13, 2005||Lefthand Networks, Inc.||Distributed Network Storage System With Virtualization|
|JP2001022614A||Title not available|
|JP2003316522A||Title not available|
|U.S. Classification||711/165, 711/162, 711/111, 711/161, 711/6, 711/114|
|International Classification||G06F12/00, G06F12/02|
|Cooperative Classification||Y10S707/99953, Y10S707/99955, G06F3/0647, G06F3/067, G06F3/0605, G06F3/0685, G06F3/0664, H04L67/1097|
|European Classification||G06F3/06A6D, G06F3/06A4H2, G06F3/06A2A2|